System and Method for Supervising Automated Imaging Modality Movement in a Medical Scene

Abstract

This invention is related to a supervision system that monitors automated movements performed by components of an medical imaging modality, in order to ensure that the moving components behave as expected, while identifying at the same time potential conflicts with (alien) objects or persons in the medical scene. The invention is based on the analysis of differences between measured distance data obtained by a detector that is mounted on the medical imaging modality, and a calculated virtual model of the geometric state of the modality components in the medical scene.

Claims

1. A system to supervise an automated movement of at least one movable component of an imaging modality within a medical scene to prevent collision between said movable component and other objects within said medical scene, said movable component providing a data stream comprising at least one positional parameter determining said movable components' geometric state in the medical scene, the system comprising, (i) an interface adapted to read out said positional parameter from said data stream from said movable component, (ii) a memory storing: (a) dynamic model data representing geometric knowledge of said at least one movable component in the medical scene as a function of said positional parameter of said movable component in said data stream, (b) stationary model data representing geometric knowledge on all non-movable objects in the medical scene, and (c) a medical scene map, (iii) at least one distance detector fixed at a known position in said medical scene, said known position is stored in said medical scene map, said distance detector providing at least one measured distance between said distance detector and an observed object along a detection axis of said distance detector, and (iv) a processor configured to: (a) calculate said medical scene map from said dynamic model data in combination with said at least one positional parameter, and from said stationary model data, (b) calculate a calculated distance between said stored position of said distance detector in said medical scene map and a first object along said detection axis of said distance detector in said medical scene map, and (c) compare said at least one measured distance with said calculated distance, such that when the difference between said at least one measured distance and said calculated distance exceeds a threshold value, a trigger signal is sent to said imaging modality.

2. The system according to claim 1, wherein said at least one distance detector is a 3D- or depth-camera.

3. The system according to claim 1, wherein said at least one distance detector is a mm-wave radar with an array of transmitting and receiving antennas.

4. The system according to claim 1, wherein said at least one distance detector is a phased array mm-wave radar.

5. The system according to claim 1, wherein said at least one distance detector is an array of ultrasound transceivers including phased arrays.

6. The system according to claim 1, wherein said other objects comprise movable and non-movable imaging modality components within said medical scene.

7. The system according to claim 1, wherein said medical scene map is two-dimensional.

8. The system according to claim 1, wherein said medical scene map is three-dimensional.

9. The system according to claim 1, wherein said positional parameter is an angle between a joint connecting said movable component and another component.

10. The system according to claim 1, wherein said positional parameter is a distance along which the movable component can extend in a predetermined direction.

11. The system according to claim 1, wherein said known position of said distance detector is preferably on a surface of said movable component.

12. A method to supervise automated movement of at least one movable component of an imaging modality within a medical scene to prevent collision between said movable component and other objects within said medical scene, said movable component providing a data stream comprising at least one positional parameter determining said movable components' geometric state in the medical scene, the method comprising the steps of, reading out said positional parameter from said data stream from said movable component, storing in a memory dynamic model data representing geometric knowledge of said at least one movable component in the medical scene as a function of said positional parameter of said movable component in said data stream, storing in a memory stationary model data representing geometric knowledge on all non-movable objects in the medical scene, measuring a distance between a distance detector and an observed object along a detection axis of said distance detector, said distance detector being fixed at a known position in said medical scene, and storing said known position in said medical scene map, calculating said medical scene map from said dynamic model data in combination with said at least one positional parameter, and from said stationary model data, and storing said medical scene map in a memory, calculating a calculated distance between said stored position of said distance detector in said medical scene map and a first object along said detection axis of said distance detector in said medical scene map, and comparing said at least one measured distance with said calculated distance, and sending a trigger signal to said imaging modality when the difference between said at least one measured distance and said calculated distance exceeds a threshold value.

Description

DESCRIPTION OF EMBODIMENTS

[0040] In the following detailed description, reference is made in sufficient detail to the above referenced principles, allowing those skilled in the art to practice the embodiments explained below.

[0041] As explained above, the system comprises among others, a processor and a memory. The memory comprises the model data (including the stationary and dynamic model data) that can be processed by the processor to produce or calculate a medical scene map. The calculation of the medical scene map is performed by the processor based on the model data stored in the memory for the set of data stream readings representing the positional parameters, and that are read from the interfaces from each movable component. The particular medical scene map that is produced for the actual positional parameters from the imaging modality components thus represents a virtual representation of the geometric state of the imaging modality, that is stored in the memory. This medical scene map is thus a virtual “copy” in memory of the actual geometric state of the imaging modality, and is based on the actual readings of the positional parameters for all components from the data streams.

[0042] The supervision function for the automated movement by an imaging modality is conceived as a continuous process whereby a calculated distance between a point on a surface of an object facing a distance detector and the detector mounting position is compared with the measured distance readings from said detector in the direction of said point on said surface of an object.

[0043] Since the mounting location of the distance detector on a surface of an object or modality component in the medical scene is known, it has to be comprised in the model data. The medical scene map, which then also comprises the location and orientation of the distance detector, will allow the calculation of any distance (calculated distance) between this distance detector mounting position and any point on an object surface.

[0044] The supervision process executed by the processor of the system thus simultaneously collects the distance measurement data (measured distance) and all positional parameters for a certain geometric state of the imaging modality. The positional parameter data are then used as input data for the model data and processed by the processor into a medical scene map. Based on this medical scene map, the calculated distances corresponding to the measured distances by the distance detector are calculated and compared with the measured distances. This process is continuously performed, and this at least while the modality components are moving.

[0045] As explained before, the medical scene map is a virtual representation of the actual state of the modality and its surroundings. Therefore, in order to assure that the virtual representation (in memory) matches the actual geometric state of the imaging modality, a distance detector can be positioned at a known location and orientation within the medical scene to validate the positions and orientations of observed or detected objects that are within its field-of-view or detection range. In case that the distance detector measures a perpendicular distance between its detector surface and a next object, this measured distance should match the equivalent calculated distance between the detector position stored in the medical scene map and the object from the medical scene map.

[0046] The comparison of the calculated distances versus the measured distances allow the evaluation of the proper functioning of the system, as each movement of the movable modality components will generate different measurement results for carefully positioned distance detectors. These different measurement results should in each case match the corresponding calculated distances from the medical scene map. Discrepancies between the measured distances and the calculated distances should be carefully compared within the tolerances of the distance measurement detectors. In case that structural deviations would occur, the supervision system should trigger the imaging modality allowing it to react appropriately. The trigger signal can then for instance alert a user, or for instance prevent further automated movement, or alternatively correct the movement course of the movable component. Structural deviations can occur when the assumed distance detector position and orientation does not match the actual detector position and orientation. In this case the distance detector position should be readjusted and calibrated to again match the measurement direction and position of the detector in the model data.

[0047] For better accuracy of the results, it is preferable that more than one distance is measured and compared with the calculated ones from the medical scene map. It is therefore preferable that more than one distance detector would be installed, and this preferably at carefully selected positions.

[0048] Further improvement of the technique can be achieved, when more distances can be measured at once for comparison when using detectors that can simultaneously measure multiple distances in different directions from the same position. The use of a 3D-camera or depth-camera, as well as depth scanners based on radar, microwaves or alike may be envisaged. In the case of the use of a depth-camera, the pixel values in an acquired depth image represent a value for the respective absolute distance between the detector and an object along the path of the respective pixel. This means that each depth image pixel can be considered as a distance measurement. Based on the medical scene map, a virtual depth-image can in principle be constructed providing a virtual depth view corresponding to the same acquisition geometry as the depth-camera. The virtual depth-image can then be compared with the acquired depth-image in order to validate the position and view of all involved objects or components.

[0049] Better results may also be achieved when carefully selecting the locations of the distance detectors. Ideally, the measured distances should vary when the modality state changes. In other words, it is preferable that the distance detector would be mounted on a movable component of the imaging modality, as this ensures that any movement of the modality component on which the detector is fixed would cause a different distance measurement. Moreover, the detection axis of the distance measurement should ideally be oriented towards other movable imaging modality components, such that again any movement of said movable components would cause changes in the measured distance by the distance detector. If the detector(s) are mounted on the surfaces of non-movable objects such as non-movable modality components or walls at least one moveable component shall be present within the field of view to again generate changing distances.

[0050] As explained above, the comparison of the expected distances versus the measured distances allow the evaluation of the proper functioning of the system, but at the same time could also indicate the presence of foreign objects or persons in the medical scene. In case that the measured distance from a distance detector does not match the anticipated and calculated distance of the medical scene map; there could be two causes: either the supervision system does not function properly as explained above, or the discrepancy is caused by the presence of an unidentified object (or person) that is not present in the medical scene map.

[0051] An unidentified object should be identified as such by the system in order to release the system for further use. This could be achieved in many ways; for instance, an operator could overrule the supervision system by confirming that the detected object is indeed an object that is not registered in the medical scene map. Or the supervision system could identify the object by confirming its location and presence by measuring distances with different detectors (and thus from different viewing angles). Another method is to identify and use the known background of the scene in case the unknown object does not fill the complete field of view and is moving. A further alternative is to observe the scene for a longer time period with the detector moving; the (static) unknown object should move consistently in the depth image.

[0052] The supervision system could bring automated modality movement even one step further by using the information about the presence, location and moving direction of unidentified objects to adjust the course of an automated modality movement during its movement. The supervision system should then detect the presence, position and moving direction of an object or person at a sufficiently high rate in order to allow to adjust the automated movements on-the-fly.